Prevention of CCAAT/Enhancer-binding Protein β DNA Binding by Hypoxia during Adipogenesis*

Upon exposure to adipogenesis-inducing hormones, confluent 3T3-L1 preadipocytes express C/EBPβ (CCAAT/enhancer binding protein β). Early induced C/EBPβ is inactive but, after a lag period, acquires its DNA-binding capability by sequential phosphorylation. During this period, preadipocytes pass the G1/S checkpoint synchronously. Thr188 of C/EBPβ is phosphorylated initially to prime the factor for subsequent phosphorylation at Ser184 or Thr179 by GSK3β, which translocates into the nuclei during the G1/S transition. Many events take place during the G1/S transition, including reduction in p27Kip1 protein levels, retinoblastoma (Rb) phosphorylation, GSK3β nuclear translocation, and C/EBPβ binding to target promoters. During hypoxia, hypoxia-inducible factor-1α (HIF-1α) is stabilized, thus maintaining expression of p27Kip1, which inhibits Rb phosphorylation. Even under normoxic conditions, constitutive expression of p27Kip1 blocks the nuclear translocation of GSK3β and DNA binding capability of C/EBPβ. Hypoxia also blocks nuclear translocation of GSK3β and DNA binding capability of C/EBPβ in HIF-1α knockdown 3T3-L1 cells that fail to induce p27Kip1. Nonetheless, under hypoxia, these cells can block Rb phosphorylation and the G1/S transition. Altogether, these findings suggest that hypoxia prevents the nuclear translocation of GSK3β and the DNA binding capability of C/EBPβ by blocking the G1/S transition through HIF-1α-dependent induction of p27Kip1 and an HIF-1α/p27-independent mechanism.

The oxygen pressure in normoxic tissue has been estimated to be 2-9% (14.4 -64.8 mm Hg). As such, physiological normoxia represents a much lower oxygen concentration than that of air, which is 20% O 2 . Insufficient blood circulation limits the delivery of oxygen and causes hypoxia, defined as 1-3% O 2 (1,2). To maintain oxygen homeostasis during hypoxic stress, cells activate many signaling molecules, including hypoxia-inducible factor (HIF), 3 endoplasmic reticulum stress response, and the environment-sensing mammalian target of rapamycin (3).
HIF plays a major role in adaptive responses to hypoxia, including erythropoiesis, angiogenesis, vasodilation, and anaerobic metabolic changes by inducing such target genes encoding erythropoietin, vascular endothelial growth factor, and inducible nitric-oxide synthase as well as those of many glycolytic enzymes. HIF, which belongs to the basic helix-loop-helix-Per-Arnt-Sim (PAS) family, is a heterodimeric transcription factor consisting of two subunits, ␣ and ␤. The first HIF-␣ isoform, HIF-1␣, was identified by affinity purification, whereas HIF-2␣/EPAS-1 was discovered in a homology search. Both isoforms heterodimerize with HIF-1␤. In normoxia, HIF-1␣ becomes ubiquitinated and rapidly degraded. A ubiquitin E3 ligase, von Hippel-Lindau protein, recognizes and binds hydroxylated proline residues in HIF-1␣. Proline hydroxylation of HIF-1␣ is catalyzed by hydroxylases specific to the factor that consume oxygen, ␣-ketoglutarate, vitamin C, and ferrous iron (Fe 2ϩ ) in the reaction (4). These cofactors are also involved in the inhibition of HIF-1␣ transcriptional activity by the asparaginyl hydroxylase factor-inhibiting HIF-1␣ (5).
Hypoxia occurs during embryogenesis, organogenesis, and in growing tumors prior to vascularization. Local hypoxic microenvironments form specific niches that modulate cell proliferation and differentiation. Recent studies have demonstrated that hypoxia regulates molecules involved in stem cell differentiation, such as Notch, ␤-catenin, Oct4, c-Myc, and Stra13/DEC1 (6 -10). Furthermore, hypoxia has been shown to inhibit adipogenesis by repressing the transcription factor, peroxisome proliferator-activated receptor-␥2 (PPAR␥2), which initiates and maintains this biological event by inducing adipocyte-specific genes. Expression of PPAR␥2 as well as C/EBP␣ is repressed by HIF-1 through induction of the basic helix-loophelix transcription repressor Stra13/DEC1 (10). PPAR␥2 and C/EBP␣ are induced by adipogenic transcription factors, including C/EBP␤, C/EBP␦, and C/EBP␣, which are activated sequentially by adipogenesis-inducing hormones (11). The C/EBP family is widely expressed, and its members play critical roles in regulating energy metabolism, inflammation, hematopoiesis, cellular proliferation, and differentiation. Despite this, C/EBPs are not expressed in 3T3-L1 preadipocytes in the absence of adipogenesis-inducing hormones (12)(13)(14)(15).
2-6 h following hormonal induction, growth-arrested 3T3-L1 preadipocytes express C/EBP␤. Early induced C/EBP␤ is inactive; however, 8 -12 h postinduction, this factor exhibits DNA binding activity and transactivation. During this lag period, the level of cell cycle-dependent kinase inhibitor p27 Kip1 decreases, and preadipocytes progress synchronously through the G 1 /S checkpoint and enter mitosis. Meanwhile, C/EBP␤ undergoes sequential phosphorylation at Thr 188 and Ser 184 (or Thr 179 ), leading to conformational changes that enable dimerization and DNA binding (16). Thr 188 is phosphorylated by p42/p44 extracellular signal-regulated kinase (ERK) and CDK2/cyclin A during the G 1 and S phases, respectively (17). Phosphorylation at this site primes C/EBP␤ for subsequent phosphorylation at Ser 184 or Thr 179 by GSK3␤, which translocates to the nucleus at the onset of S-phase.
Many reports have suggested that C/EBP␤ function is coordinated with the cell cycle (18 -20). During the G 1 /S transition, many events take place, including reduction in p27 Kip1 levels, retinoblastoma (Rb) phosphorylation, GSK3␤ nuclear translocation, dual phosphorylation of C/EBP␤, and C/EBP␤ binding to the PPAR␥2 and C/EBP␣ promoters. Despite this knowledge, the cause-and-effect relationship among these events is unclear. Here, through the use HIF-1␣ knockdown 3T3-L1 cells, we demonstrate that hypoxia inhibits these events, which can be dissected into at least two independent processes that are distinguishable by HIF-1␣.
Cell Culture and Adipogenesis of 3T3-L1 Cells-3T3-L1 (American Type Culture Collection, catalogue number CL-173) preadipocytes were maintained in Dulbecco's modified Eagle's medium containing 10% (v/v) bovine calf serum. Postconfluent 3T3-L1 cells were induced to differentiate by the addition of a standard mixture (MDI) composed of 0.5 mM 3-isobutyl-1-methylxantine, 1 M dexamethasone, and 5 g/ml insulin in 10% fetal bovine serum for the first 48 h. The cells were then cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum and 5 g/ml insulin in a humidified atmosphere of 95% air and 5% CO 2 at 37°C. The medium was replaced every 2 days. Cells were exposed to hypoxic conditions (1% O 2 ) by incubation in an anaerobic incubator (model 1029, Forma Scientific, Inc.) in an atmosphere of 5% CO 2 , 10% H 2 , and 85% N 2 at 37°C.
Oil Red O Staining-Confluent 3T3-L1 cells were induced to differentiate for the indicated number of days. Cells were washed once with PBS, fixed with 10% formaldehyde in PBS for 1 h at room temperature, and then washed again with PBS. Oil Red O (0.5% in isopropyl alcohol) was diluted with distilled water at a ratio of 3:2 (v/v) and incubated with the fixed cells for 30 min at 37°C. Cells were washed with distilled water for 10 min at 37°C. Stained lipid droplets in the cells were visualized by light microscopy.
Transfection and Retroviral Infection-To generate 3T3-L1 cells that constitutively express HIF-1␣ and p27 Kip1 , we used a retroviral expression system (BD Biosciences). Mouse p27 Kip1 cDNA (GenBank TM accession number NM_009875) was subcloned into the retroviral vector pWZL-IRES-neo encoding Geneticin resistance, which was a generous gift from Garry Nolan (Stanford University School of Medicine, Stanford, CA). HEK293-based packaging cells (BD AmphoPack TM -293 cell line) were transfected with pWZL-IRES-neo-p27 Kip1 . To generate control retrovirus, the packaging cells were transfected with an empty retroviral vector. 48 h post-transfection, retrovirus was collected from the medium of packaging cells (21). 3T3-L1 cells at 50% confluence were infected with virus-containing medium in the presence of 8 g/ml Polybrene according to the manufacturer's instructions. The infected 3T3-L1 cells were selected in the presence of 0.5 mg/ml G418. To generate HIF-1␣-knockdown cells, we used a retroviral vector system (BD Biosciences). Short hairpin RNA (shRNA) against HIF-1␣ (5Ј-GATCCGTGTGAGCTCACATCTTGATTT-CAAGAGAATCAAGATGTGAGCTCACATTTTTTAGAT-CTG-3Ј) was inserted into the pSIREN-RetroQ vector (BD Biosciences) according to the manufacturer's instructions to generate pSIREN-RetroQ-shHIF-1␣. The infected 3T3-L1 cells were selected in the presence of 5 g/ml puromycin. To generate control retrovirus, control shRNA provided with the kit was ligated into the pSIREN-RetroQ vector.
Preparation of Nuclear Extracts-Confluent 3T3-L1 cells were induced to differentiate for the indicated times in 100-mm culture plates. Cells were washed once with cold 10 mM HEPES, pH 7.5, and then incubated in fresh buffer on ice for 15 min. Cells were harvested with MDH buffer (3 mM MgCl 2 , 1 mM dithiothreitol, 25 mM HEPES, pH 7.5, 0.4 mM phenylmethylsulfonyl fluoride, 2 g/ml leupeptin, and 2 g/ml aprotinin) and Dounce homogenized (20 strokes) using a type B pestle on ice. Nuclei were isolated by centrifugation at 1,000 ϫ g for 5 min at 4°C, washed twice with MDHK buffer (3 mM MgCl 2 , 1 mM dithiothreitol, 25 mM HEPES, pH 7.5, 100 mM KCl, 0.4 mM phenylmethylsulfonyl fluoride, 2 g/ml leupeptin, and 2 g/ml aprotinin) and then resuspended in HDK buffer (25 mM HEPES, pH 7.5, 1 mM dithiothreitol, 400 mM KCl, 0.4 mM phenylmethylsulfonyl fluoride, 2 g/ml leupeptin, and 2 g/ml aprotinin). The extracts were incubated with gentle rocking at 4°C for 1 h and then centrifuged at 16,000 ϫ g for 1 h at 4°C. Supernatants were isolated as nuclear extracts. Protein concentrations were measured by the Bradford assay (Bio-Rad) (22).
FACS Analysis-Confluent 3T3-L1 cells were induced to differentiate for the indicated number of hours, washed once with PBS, harvested by trypsinization, and centrifuged at 1,000 ϫ g for 5 min. The cells were fixed overnight with 70% ethanol at Ϫ20°C. The fixed cells were washed with PBS containing 2 mM EDTA and then centrifuged. The pellet was resuspended in 1 ml of PI buffer (PBS containing 50 g/ml PI and 40 g/ml RNase A) and incubated at room temperature for 30 min. PIstained cells were analyzed using a BD FACSCalibur TM flow cytometer (BD Biosciences), and the data were analyzed using CellQuest (BD Biosciences).
Chromatin Immunoprecipitation (ChIP) Analysis-A ChIP assay kit was used according to the manufacturer's instructions (Millipore, Billerica, MA). 3T3-L1 preadipocytes were induced to differentiate as described. The cells were cross-linked with 1% formaldehyde at room temperature for 10 min and then washed twice with ice-cold PBS prior to harvest. Cells were resuspended in 200 l of SDS lysis buffer (50 mM Tris-HCl, pH 8.1, containing 1% SDS, 10 mM EDTA, and protease inhibitors), incubated on ice for 10 min, and then sonicated on ice to yield DNA fragments with an average length of 0.5-1 kb. After centrifugation at 13,000 ϫ g for 10 min at 4°C, the concentration of lysate was determined by spectrophotometry. Equivalent amounts of chromatin were used for immunoprecipitation. The lysates were diluted at a ratio of 1:10 with ChIP dilution buffer (20 mM Tris-HCl, pH 8.1, containing 1% Triton X-100, 2 mM EDTA, and 150 mM NaCl) and incubated with 2 g of salmon sperm DNA and 20 l of protein A/G-agarose at 4°C for 1 h. After centrifugation, 10% of the supernatant was removed as input and stored at Ϫ20°C, and the DNA-protein complexes were immunoprecipitated with 2 g of anti-C/EBP␤ antibody (H-7) or 2 g of anti-p300 antibody (N-15) (Santa Cruz Biotechnology, Inc.) at 4°C overnight. The antibodychromatin complexes were recovered by incubation with 20 l of protein A/G-agarose. Immunocomplexes were eluted with elution buffer (1% SDS and 0.1 M NaHCO 3 ), and NaCl was added to a final concentration of 200 mM prior to overnight at 65°C incubation to reverse cross-linking. Finally, DNA was extracted with phenol/chloroform, precipitated, and resuspended. Precipitated DNAs were amplified by PCR (30 -35 cycles) and analyzed by 2% agarose gel electrophoresis. PCR products were radiolabeled by inclusion of 0.25 Ci of [␣-32 P]dATP in the reaction mixture followed by PCR (20 -25 cycles). Radiolabeled PCR products were resuspended in 8% polyacrylamide, 1ϫ Tris borate-EDTA gels, dried, and exposed to x-ray film. The primers used for PCR were as follows: C/EBP binding site in C/EBP␣ promoter (GenBank TM accession number NM_007678), 5Ј-TCCCTA-GTGTTGGCTGGAAG-3Ј (forward) and 5Ј-CAGTAGGA-TGGTGCCTGCTG-3Ј (reverse); C/EBP binding site in the PPAR␥2 promoter (GenBank TM accession number S79407), 5Ј-TTCAGATGTGTGATTAGGAG-3Ј (forward) and 5Ј-AGACTTGGTACATTACAAGG-3Ј (reverse); HIF-1␣ binding site in the Stra13 promoter (GenBank TM accession number AF362845), 5Ј-aacacgtgaggctcatgt-3Ј (forward) and 5Ј-gttaaatgggagcgagtg-3Ј (reverse).
Immunofluorescence Microscopy-3T3-L1 preadipocytes (2 ϫ 10 4 ) were plated on glass coverslips in a 24-well culture plate and cultured until confluent. The cells were then induced to differentiate as described above. Cells were washed with PBS, fixed with 4% paraformaldehyde in PBS for 20 min at room temperature, permeabilized with 0.4% Triton X-100 in PBS for 30 min, and then washed with PBST buffer (PBS containing 0.05% Tween 20). The cells were blocked in PBST buffer containing 5% bovine serum albumin for 1-2 h at room temperature. The cells were incubated overnight with anti-C/EBP␤ (H-7, 1:1,000) or anti-p27 (C-19, 1:200) antibodies in PBST buffer containing 5% bovine serum albumin and washed with PBST buffer. The cells were incubated with secondary antibody conjugated to either AlexaFluor 546 or AlexaFluor 488 (Invitrogen) for 1 h at room temperature. Cells mounted onto coverslips were incubated with DAPI (Invitrogen) for 5 min and then washed with PBST buffer. After the coverslips were mounted on slides, stained cells were observed under a Zeiss LSM510 inverted confocal microscope according to the manufacturer's instructions.

Hypoxia Represses PPAR␥ and C/EBP␣ but Not C/EBP␤-To
investigate the effects of hypoxia on adipogenesis, 3T3-L1 preadipocytes were induced to differentiate into mature adipocytes by hormone treatment. Staining with Oil Red O revealed lipid droplets in 3T3-L1 cells after treatment for 8 days, indicating that the preadipocytes had differentiated under normal conditions. Lipid droplets, however, were not observed in 3T3-L1 cells treated under hypoxic conditions (Fig. 1A). Northern and Western analyses showed that during adipogenesis, PPAR␥2 and C/EBP␣ and their target, adiponectin, were induced under normoxic conditions but not under hypoxic conditions. In contrast, HIF-1␣ and its target, vascular endothelial growth factor, were induced under hypoxia ( Fig. 1, B-D) (25). Because PPAR␥2 and C/EBP␣ were reported to be induced by C/EBP␤, we tested whether hypoxia also abrogated this induction. Interestingly, the expression level of C/EBP␤ was unaffected by hypoxic conditions (Fig. 1, D and E).
Hypoxia Inhibits the DNA Binding Ability of C/EBP␤-Next we investigated whether hypoxia affects C/EBP␤ activity. Although this factor is induced early, within 4 h of the start of differentiation, it is not immediately active. C/EBP␤ acquires its DNA binding ability 8 -12 h postinduction (18). ChIP analyses using an anti-C/EBP␤ antibody revealed that C/EBP␤ binds the C/EBP␣ and PPAR␥2 promoters in vivo 12 h after induction under normoxic but not hypoxic conditions (Fig. 2, A and B).
To confirm this finding, we tested whether p300, a C/EBP␤ coactivator, is also recruited to the promoters of C/EBP␤ target genes. ChIP analyses demonstrated that p300 also occupies the endogenous promoter regions of C/EBP␣ and PPAR␥2 at 12-24 h after hormonal induction. Like C/EBP␤, p300 does not occupy these promoter regions under hypoxic conditions. However, this coactivator was found bound to the promoter of Stra13/DEC1, which is induced by HIF-1/p300 in response to hypoxia (Fig. 2C). Interestingly, electrophoretic mobility shift assay showed a reduction in the number of nuclear proteins bound to probes containing C/EBP binding sites (supplemental Fig. 1A).
As preadipocytes enter S-phase, C/EBP␤ acquires DNA binding ability and localizes to centromeres by binding to C/EBP consensus-binding sites within centromeric satellite DNA. By binding to centromeric DNA, C/EBP␤ becomes associated with punctate nuclear components (19). Consistent with this, immunofluorescence revealed that 20 h after hormone induction, C/EBP␤ formed a punctate pattern, demonstrating that C/EBP␤ binds to pericentromeric satellite DNA as preadipocytes enter S-phase (Fig. 2D) (18,26). In contrast, C/EBP␤ was distributed diffusely within nuclei 4 h post-treatment. This diffuse staining pattern was also observed even at 20 h after induction under hypoxic conditions. These findings indicate that hypoxia prevents C/EBP␤ DNA binding during adipocyte differentiation.
We also tested the expression level of C/EBP␦, the other early induced C/EBP family during adipogenesis, and its localization in hypoxic conditions. Hypoxia does not affect C/EBP␦ expression and localization (data not shown). The C/EBP homologue CHOP10, also known as GADD153 or C/EBP, inhibits the DNA binding capacity of C/EBP␤ through heterodimerization with the factor. The expression level of CHOP10 corre-lates inversely with adipogenesis (27). Reports have demonstrated that anoxia (0% O 2 ) and endoplasmic reticulum stress increase CHOP10 expression (28). Western analysis revealed that CHOP10 levels decrease, whereas that of C/EBP␤ increases (Fig. 2E). However, hypoxic treatment did not affect CHOP10 protein levels during adipogenesis, suggesting that the inability of C/EBP␤ to bind DNA is not due to an increase in CHOP10 during hypoxia.
Hypoxia Blocks the Mitotic Clonal Expansion (MCE) of Preadipocytes-Although C/EBP␤ gains its DNA binding ability as preadipocytes progress through the G 1 /S transition, the cause-and-effect relationship between these two events is unclear. FACS analyses showed that 3T3-L1 cells undergo the G 1 /S transition 12-18 h after induction. However, hypoxia leads to G 1 -phase arrest (Fig. 3A). Cell-to-cell contact in confluent 3T3-L1 cells abrogates cell cycle progression by inducing the cyclin-dependent kinase inhibitor p27 Kip1 , which suppresses the activity of CDK4 (cyclin-dependent kinase 4). CDK4 phosphorylates Rb, which is a prerequisite for the G 1 /S transition. In early G 1 , corresponding to 1-4 h postinduction (Fig. 3B), p27 Kip1 protein is detected, and Rb is hypophosphorylated (29). However, 12-24 h after induction, p27 Kip1 is diminished, leading to phosphorylation of Rb and resumption of the cell cycle (30,31). In contrast, under hypoxic conditions, p27 Kip1 levels remain high because this protein is induced by HIF-1␣ (32) (see also Fig. 3B). Thus, hypoxia maintains p27 Kip1 at high levels, keeping Rb hypophosphorylated regardless of adipogenesis-inducing hormone treatment (MDI) (Fig. 3B) (33).
block mitotic clonal expansion, thus preventing C/EBP␤ activation. Therefore, hypoxia can block 3T3-L1 cell adipogenesis by promoting HIF-1␣-dependent induction of p27 Kip1 . G 1 /S Transition Does Not Occur in Hypoxic HIF-1␣ Knockdown 3T3-L1 Cells-We further investigated whether HIF-1␣ is indispensable for hypoxic inhibition of C/EBP␤ DNA binding ability. To perform these experiments, HIF-1␣-deficient 3T3-L1 cells were generated by retroviral infection of shRNA specific against mouse HIF-1␣. HIF-1␣ was stable during hypoxia in 3T3-L1 cells infected with retrovirus encoding control shRNA but not HIF-1␣ shRNA (Fig. 4A). Moreover, hypoxia induced p27 Kip1 in 3T3-L1 cells infected with control shRNA but not in HIF-1␣ shRNA. This is consistent with a previous report showing the failure of hypoxia to induce p27 Kip1 in HIF-1␣ knock-out mouse embryonic fibroblasts (32). Nonetheless, Rb was not phosphorylated under hypoxic conditions even in 3T3-L1 cells where neither HIF-1␣ nor p27 Kip1 was induced (Fig. 4A, lanes 8 and 10, and supplemental Fig. 2A). FACS analyses showed that hypoxia arrests HIF-1␣ knockdown 3T3-L1 cells at G 1 -phase, similar to control cells (Fig. 4B). These findings suggest that hypoxia inhibits Rb phosphorylation and G 1 /S transition in the absence of p27 Kip1 in differentiating 3T3-L1 cells. Gardner et al. (33) demonstrated that, in p27 Kip1 -deficient murine fibroblast cells, hypoxia neither inhibits Rb phosphorylation nor causes G 1 -phase arrest, indicating that HIF-1␣ induction of p27 Kip1 is the primary pathway by which Rb hypophosphorylation and G 1 arrest occur in response to hypoxia. However, this pathway is not unique to differentiating 3T3-L1 cells. Our results indicate that hypoxia can block the G 1 /S transition and Rb phosphorylation in a HIF-1␣-independent manner, especially in differentiating 3T3-L1 cells.
Furthermore, induction of PPAR␥ and C/EBP␣ and lipid droplet formation in response to adipogenesis-inducing hormones during normoxia were unaffected by HIF-1␣ knockdown (Fig. 5, A and B). These findings confirmed that infected preadipocytes retain their ability to differentiate into mature adipocytes regardless of the presence of HIF-1␣. In contrast, hypoxia blocks hormonal induction of PPAR␥ and C/EBP␣ and lipid droplet formation in the absence of HIF-1␣ (Fig. 5, A and  B). ChIP analyses further verified these results because C/EBP␤ binding to the PPAR␥2 and C/EBP␣ promoters was abrogated by hypoxia even in 3T3-L1 cells lacking HIF-1␣ (Fig. 5C). Immunofluorescence demonstrated that hypoxia prevents C/EBP␤ from acquiring a punctate pattern in these cells (Fig.  5D). These findings indicated that hypoxia prevented C/EBP␤ from acquiring DNA binding ability not only by HIF-1␣-dependent induction of p27 Kip1 but also by a HIF-1␣-independent pathway.
Dual Phosphorylation of C/EBP␤ by both ERK and GSK3␤-Recent studies have suggested that dual phosphorylation of C/EBP␤ is required for this factor to acquire DNA binding ability (16,17,37). Both ERK, which is activated early in the G 1 -phase, and CDK2/cyclin A, which is activated at the G 1 /S FIGURE 2. DNA binding activity and heterochromatin localization of C/EBP␤. Postconfluent 3T3-L1 cells were treated with MDI for the indicated times and then fixed and harvested for ChIP analysis. A, ChIP analysis of C/EBP␤ binding to the PPAR␥2 promoter in vivo. Using anti-C/EBP␤ antibody, protein-genomic DNA complexes were immunoprecipitated (IP) and analyzed by 30 cycles of PCR. Input indicates that PCR was performed using 10% of the amount of sonicated cell lysate that was used for immunoprecipitation. B, using anti-C/EBP␤ antibody, protein-genomic DNA complexes were immunoprecipitated and analyzed by 22 cycles of PCR. PCR was performed in the presence of [␣-32 P]dATP and primers for C/EBP␣ (top) or the PPAR␥2 promoter (bottom). C, ChIP analysis of p300 binding to the promoters of C/EBP␣ (top), PPAR␥2 (middle), and Stra13 (bottom) in vivo. D, confocal microscopic images of C/EBP␤ in differentiating 3T3-L1 cells. One cell in a white square was magnified and shown on the right. E, Western analysis of treated 3T3-L1 cells using anti-CHOP10, anti-C/EBP␤, anti-HIF-1␣, and anti-␣-tubulin antibodies. ␣-Tubulin was used as a loading control.
transition, maintain C/EBP␤ phosphorylation at Thr 188 throughout the G 1 and S phases. Phosphorylation of this residue is required for subsequent phosphorylation of Thr 179 or Ser 184 by GSK3␤ (16). By using in vitro phosphorylation of C/EBP␤ and an electrophoretic mobility shift assay, we confirmed that both ERK and GSK3␤ are required for the dual phosphorylation of C/EBP␤ and acquisition of its DNA binding ability (supplemental Fig. 1, B and C). To determine whether hypoxia blocks this sequential process, we investigated ERK activation by Western blot analysis. The level of phosphorylated ERK increased within 1 h of induction and diminished within 12 h. ERK remained phosphorylated even under hypoxic conditions (Fig. 6A). Therefore, hypoxia does not block the activation of ERK. Western blot analysis of C/EBP␤ showed that Thr 188 phosphorylation was sustained 12-24 h after induction (Fig. 6B). This antibody specifically recognizes phosphorylated Thr 188 on C/EBP␤ (supplemental Fig. 1D) (17). These data indicate that Thr 188 is not involved in the inability of C/EBP␤ to bind DNA during hypoxia.
As adipogenesis proceeds, the amount of GSK3␤ in the nucleus increases gradually before terminal differentiation. Tang et al. suggested that, during the G 1 /S transition, GSK3␤ translocates into the nucleus, where it phosphorylates other nuclear proteins, including C/EBP␤ and cyclin D1 (31,38). Our data demonstrate that hypoxia blocks GSK3␤ nuclear translo-cation during adipogenesis (Fig. 6C), thereby preventing phosphorylation of C/EBP␤ at Thr 179 and/or Ser 184 . In addition, immunoblotting using an antibody specific for phosphorylated Ser 9 of GSK3␤ revealed that hypoxia increases the level of inactive GSK3␤ (Fig. 6D).
The Role of GSK3␤ and ERK in C/EBP␤ DNA Binding-We further investigated the biochemical mechanism involved in hypoxia-induced inhibition of C/EBP␤ DNA binding. Immunofluorescence showed that treatment of cells with the GSK3␤ inhibitors LiCl and 6-bromoindirubin-3Ј-oxime (BIO) for 8 or 20 h before harvest abrogated the punctate distribution of C/EBP␤ (Fig. 7A). Likewise, treatment with U0126, an inhibitor of mitogen-activated protein kinase/extracellular signal-regulated kinase kinase (MEK), the upstream kinase of ERK, for 20 or 12 h (20 h and Early 12 h indicated in Fig. 7) at the beginning of MDI treatment produced the same effect. Later exposures (i.e. 2 and 8 h (Late 2 h and Late 8 h indicated in Fig. 7)) of U0126 produced a punctate staining pattern, confirming that ERK is activated only during early G 1 -phase, which occurs ϳ1-12 h after MDI treatment (Fig. 6A). Moreover, shorter exposures of hypoxia or the GSK3␤ inhibitors for 2 h (Late 2 h indicated in Fig. 7) at the end of MDI treatment also yielded a punctate C/EBP␤ distribution, indicating that GSK3␤ activity is required during the G 1 /S transition, which occurs approximately 12-18 h after MDI treatment. The results from ChIP analyses shown in Fig. 7B confirm FIGURE 3. Mitotic clonal expansion of 3T3-L1 cells during adipogenesis. Postconfluent 3T3-L1 cells were induced to differentiate by treatment with MDI under normoxic or hypoxic conditions. A, flow cytometry of cells under normoxic (N) and hypoxic (H) conditions. B, Western analysis of treated 3T3-L1 cells was performed with anti-Rb, anti-p27, and anti-14-3-3␥-antibodies. 14-3-3␥ was used as a loading control. C, effects of retroviral expression of p27 Kip1 in 3T3-L1 cells. As a control, 3T3-L1 cells were infected with retrovirus encoding an empty vector (EV). The selected 3T3-L1 cells were treated with MDI. Eight days after differentiation, 3T3-L1 cells were stained with Oil Red O as described. The levels of PPAR␥ and C/EBP␣ protein were determined by Western analysis. 14-3-3␥ was used as a loading control. D, the selected 3T3-L1 cells were treated with MDI for 20 h under normoxia. The protein levels of p27, C/EBP␤, and Rb were determined by Western blot. E, the selected 3T3-L1 cells were induced to differentiate for the indicated times, fixed with 70% ethanol, and stained with PI. The PI-stained cells were analyzed by flow cytometry. F, confocal microscopic images of C/EBP␤ in differentiating 3T3-L1 cells. The infected 3T3-L1 cells were treated with MDI for 20 h under normoxia. The stained C/EBP␤ in 3T3-L1 cells was observed with a Zeiss LSM510 inverted confocal microscope. One cell in a white square was magnified and shown on the right. that both U0126 and BIO prevent C/EBP␤ binding to the promoters of C/EBP␣ and PPAR␥2, suggesting that ERK and GSK-3␤ are required for C/EBP␤ to acquire its DNA binding ability.
GSK-3␤ also phosphorylates ␤-catenin, leading to its degradation. We confirmed that LiCl and BIO treatment increased ␤-catenin nuclear translocation by blocking GSK-3␤ activity. MDI treatment causes GSK-3␤ levels to increase within the nucleus and ␤-catenin to decrease. However, during hypoxia, GSK-3␤ levels within the nucleus did not increase, allowing ␤-catenin levels to also remain unaffected in MDI-treated 3T3-L1 cells (Fig. 7C). The finding that Wnt inhibits adipogen-esis via the ␤-catenin/T-cell factor (Tcf)-dependent pathway suggests that, during hypoxia, increased levels of ␤-catenin contribute to the hypoxic repression of PPAR␥2 (39). Different from hypoxia, GSK3␤ inhibitors did not induce p27. Thus, Rb remains phosphorylated in MDI-treated 3T3-L1 cells, even in the presence of either LiCl or BIO. These results indicated that GSK3␤ activity is not required for Rb phosphorylation (Fig. 7D). Moreover, we discovered that constitutive expression of p27 in 3T3-L1 cells prevented GSK3␤ nuclear translocation in response to adipogenesisinducing hormones even under normoxia (Fig. 8A), suggesting that Rb phosphorylation and the G 1 /S transition cause GSK3␤ translocation into the nucleus.

DISCUSSION
Because C/EBP␤ is sequentially phosphorylated during the G 1 and S phases, other check points for mitosis are also expected to regulate the dual phosphorylation of C/EBP␤. Protein kinases that catalyze the phosphorylation of C/EBP␤ are tightly regulated by the cell cycle. ERK, CDK2/cyclin A, and GSK3␤ are activated during G 1 and the G 1 /S transition. GSK3␤ is localized predominantly within the cytoplasm during early G 1 -phase (2-8 h), but a significant fraction of GSK3␤ translocates to the nucleus during the G1/S transition (Fig. 6C). Because Thr 188 of C/EB␤ remained phosphorylated even during hypoxia, hypoxia inhibits the DNA binding ability of C/EBP␤ by a mechanism that does not involve this residue. Results from this study suggest that hypoxia prevents C/EBP␤ from acquiring DNA binding ability by blocking GSK3␤ nuclear translocation. Our findings that (i) GSK3␤ failed to translocate into the nucleus when Rb was hypophos- phorylated and when 3T3-L1 cells were arrested at G 1 -phase and (ii) GSK3␤ inhibitors did not block Rb phosphorylation imply that hypoxia blocks nuclear translocation of GSK3␤ by blocking G 1 /S transition but not that hypoxic inhibition of GSK3␤ nuclear translocation causes hypophosphorylation of Rb and G 1 arrest. However, it is still unclear how GSK3␤ acquires the ability to enter the nucleus during the G 1 /S transition. Further investigation is necessary to examine the possibility that hypoxia directly inhibits GSK3␤ nuclear translocation.
Hypoxia prevents most cells from traversing the G 1 /S checkpoint through hypophosphorylation of the Rb protein by inducing p27 Kip1 in a HIF-1␣-dependent manner. Goda et al. (32) demonstrated that in murine embryonic fibroblasts (MEFs) and splenic B lymphocytes deficient in HIF-1␣, hypoxia fails to block the G 1 /S transition and Rb phosphorylation, suggesting that HIF-1␣ is indispensable for these events. Gardner et al. demonstrated that in p27 Kip1 -deficient MEFs, hypoxia neither inhibits Rb phosphorylation nor causes G 1 arrest. However, in p21 Cip1 -deficient MEFs, hypoxia causes the fibroblasts to arrest in the G 1 -phase FIGURE 5. Effect of HIF-1␣ knockdown on the DNA binding ability of C/EBP␤ during adipogenesis. A, after a 6-day treatment with MDI, Western analysis was performed using anti-PPAR␥, anti-C/EBP␣, anti-HIF-1␣, and anti-14-3-3␥ antibodies. 14-3-3␥ was used as a loading control. B, after 8 days, 3T3-L1 cells were stained with Oil Red O. C, hormone-treated 3T3-L1 cells were fixed and harvested for ChIP analysis using primers against the PPAR␥2 and C/EBP␣ promoters. Input indicates that PCR was performed using 10% the amount of sonicated cell lysate that was used for immunoprecipitation (IP). D, 3T3-L1 cells were carried out on glass coverslips and differentiated for 20 h under normoxic or hypoxic conditions. Confocal microscopic images of cellular C/EBP␤ were observed using a Zeiss LSM510 microscope. One cell in a white square was magnified and shown in the upper panel. N, normoxia; H, hypoxia. . C, nuclear extracts were prepared from the hormone-treated 3T3-L1 cells as described under "Experimental Procedures." Using 10 g of nuclear extracts, Western analysis was performed with anti-GSK3␤, anti-C/EBP␤, anti-HIF-1␣, or anti-lamin A/C antibodies. Lamin A/C was used as a loading control for the nuclear extracts. D, Western analysis on whole cell extracts was performed using anti-phospho-Ser 9 -GSK3␤, anti-GSK3␤, anti-HIF-1␣, or anti-14-3-3␥ antibodies. 14-3-3␥ was used as a loading control. (33). These findings suggest that HIF-1␣-induced p27 Kip1 is a unique cell cycle inhibitor that mediates Rb hypophosphorylation and G 1 arrest in fibroblasts in response to hypoxic stress. Consistent with this, we observed that p27 Kip1 , but not p21 Cip1 , was induced by hypoxia in 3T3-L1 cells (supplemental Fig. 2B). We also showed that 3T3-L1 lacking HIF-1␣ failed to induce p27 Kip1 in response to hypoxia. Nevertheless, in hypoxic HIF-1␣-knockdown preadipocytes, Rb remained hypophosphorylated even in the absence of p27 Kip1 induction. This suggests that, particularly when under the control of adipogenesis-inducing hormones, hypoxia can block Rb phosphorylation by a mechanism distinct from the HIF-1␣-dependent induction of p27 Kip1 . By using monkey kidney epithelial cells, Krtolica et al. (40) showed that, in addition to increases in p27 Kip1 , hypoxia lowers cyclin D and E protein levels and increases Rb-directed protein phosphatase 1, thus reducing cyclin-CDK activity. Green et al. (34) showed repression of CDK2 activity and Rb hypophosphorylation in hypoxic p21 Cip1 -and p27 Kip1 -deficient (p21 Cip1Ϫ/Ϫ , p27 Kip1Ϫ/Ϫ ) MEFs, indicating that, without p21 Cip1 and p27 Kip1 , hypoxia still inhibits CDK2 activity. Hypoxia is also known to repress cdc25a, another cell cycle gene encoding a tyrosine phosphatase that maintains CDK2 activity. However, HIF-1␣ alone fails to repress this gene (41). Taken together, these findings indicate that additional factors besides HIF-1␣ and p27 Kip1 are involved in hypoxia-induced cell cycle arrest. Additional studies are required to investigate whether HIF is involved in the hypoxic regulation of cyclins, protein phosphatase 1, and Cdc25a.
We also found that hypoxia increases levels of inactive GSK3␤, which is phosphorylated on Ser 9 by the upstream kinase Akt (Fig. 6D and supplemental Fig. 2C). Therefore, hypoxia inhibits both the nuclear localization and activity of GSK3␤ in MDI-treated 3T3-L1 cells. Like other GSK3␤ inhibitors, such as LiCl and BIO, hypoxia also increases the level of nuclear ␤-catenin, a well known target of GSK3␤ (Fig. 7C). Both ␤-catenin and GSK3␤ form a complex in the cytoplasm with the common scaffold protein axin. GSK3␤ phosphorylates the closely associated ␤-catenin and triggers its degradation. In response to canonical Wnt signaling, the complex formed by GSK3␤, ␤-catenin, axin, and the anaphase-promoting complex is dissociated. Under these conditions, phosphorylation of ␤-catenin by GSK3␤ is reduced, and ␤-catenin retains greater stability and translocates into the nucleus. Canonical Wnt ligands, such as Wnt10b and Wnt3a, are known to inhibit adipogenesis by repressing both C/EBP␣ and PPAR␥ by a ␤-catenin/Tcf-dependent mechanism (39,42). A previous study using bipotential mesenchymal ST2 cells showed that Wnt signals do not alter C/EBP␤ expression (43,44). It remains to be determined, however, whether nuclear ␤-catenin is involved in inhibiting the DNA binding ability of C/EBP␤. Both conditions. The cells were also exposed to hypoxia for the indicated times. Confocal microscopic images of cellular C/EBP␤ were observed using a Zeiss LSM510 microscope. N, normoxia; H, hypoxia. B, in the presence of U0126 (U) (20 M), BIO (1 M), or DMSO (Ϫ) (0.1%, v/v), the postconfluent 3T3-L1 cells were induced to differentiate by treatment with MDI for 24 h. The treated 3T3-L1 cells were then fixed and harvested for ChIP analyses using an anti-C/EBP␤ antibody and primers against the C/EBP␣ (top) and PPAR␥2 promoters (bottom). Input indicates that PCR was performed using 10% of the amount of sonicated cell lysate that was used for immunoprecipitation. C, postconfluent 3T3-L1 cells were treated with MDI for 24 h in the presence of LiCl (20 mM) and BIO (1 M) or under hypoxic conditions. Using 10 g of the nuclear extracts of the treated 3T3-L1 cells, Western analysis was performed with anti-␤-catenin, anti-GSK3␤, anti-HIF-1␣, and anti-lamin A/C antibodies. Lamin A/C was used as a loading control for the nuclear extracts. D, Western analysis on whole cell extracts of the treated 3T3-L1 cells was performed using anti-p27, anti-Rb, anti-HIF-1␣, anti-C/EBP␤, and anti-14-3-3␥ antibodies. 14-3-3␥ was used as a loading control. hypoxia and GSK3␤ inhibitors prevent C/EBP␤ from binding to DNA. Unlike hypoxia, however, GSK3␤ inhibitors fail to inhibit the Rb phosphorylation (Fig. 7D). BIO and LiCl inhibited the catalytic activity of GSK3␤, whereas hypoxia inhibited the nuclear localization and activity of GSK3␤, presumably by blocking the G 1 /S transition and activating the Akt pathway.
Because C/EBP␤ activated in response to the G 1 /S transition triggers the expression of PPAR␥2, C/EBP␤ can mediate signals from cell cycle events to transcriptional cascades, leading to irreversible terminal differentiation. The finding that a dominant negative C/EBP␤ mutant, incapable of binding DNA, blocks MCE in 3T3-L1 preadipocytes suggests that the DNA binding ability of C/EBP␤ is also required for proceeding through MCE followed by the G 1 /S transition (19,20). In addition, the fact that active C/EBP␤ increases transcription of C/EBP␣ and PPAR␥, which were also responsible for terminating MCE, implies that C/EBP␤ is also responsible for halting MCE. Thus, C/EBP␤ is involved in limiting MCE to one or two cell divisions prior to differentiation (19,45,46). Normal embryonic fibroblasts from lung buds differentiate into adipocytes in response to hormonal induction; however, embryonic fibroblasts deficient in Rb do not, indicating that Rb is essential for adipogenesis (47). The interaction between Rb and C/EBP␤ occurred only in differentiating cells, and interaction with Rb appeared to inhibit the binding of C/EBP␤ to cognate DNA sequences in vitro (48). Forced expression of C/EBP␤ inhibited the proliferation of wild type MEFs but not of MEFs lacking all three Rb family proteins, suggesting that C/EBP␤ represses genes required for cell cycle progression through interaction with Rb proteins (49). However, it remains to be determined how the phosphorylation states of both Rb and C/EBP␤ affect these interactions.
Yun et al. (10) have demonstrated that the PPAR␥2 promoter is suppressed by Stra13/DEC1, which is induced by HIF-1␣, suggesting that HIF-1␣ and its target Stra13/DEC1 mediate hypoxic inhibition of adipogenesis after C/EBP␤-DNA interaction. Recently, Gulbagci et al. demonstrated that the Stra13/DEC1 isoform DEC2/SHARP1 interacts with C/EBP␤ and enhances the recruitment of histone deacetylase 1 and histone methyltransferase G9a to the promoter of PPAR␥2. Thus, DEC2/SHARP1 also represses adipogenesis (50). We demonstrated previously that both DEC1 and DEC2 repress the lipogenic transcription factor, SREBP-1c (sterol regulatory element binding protein-1c), by preventing binding to promoters of its targets, including the promoters of SREBP-1c and fatty acid synthase (24). These findings suggest that HIF-induced DEC1 and DEC2 are directly and indirectly involved in hypoxic inhibition of adipogenesis because SREBP-1c participates in the synthesis of a putative endogenous ligand for PPAR␥ during lipogenesis.
Hypoxia inhibits adipogenesis by blocking both the MCE and differentiation processes through (i) HIF-1-dependent induction of p27 Kip1 and (ii) HIF-1-independent inhibition of GSK3␤ and C/EBP␤ signals (Fig. 8C). By recruiting such diverse pathways, preadipocytes can arrest adipogenesis any time oxygen is limited. This idea is consistent with fact that adipose tissue is a highly vascular connective tissue. Although hypoxia has an inhibitory effect on cell differentiation, hypoxia often promotes chondrogenesis of mesenchymal stem cells (51). Consistent with this, cartilage is known to be an avascular connective tissue. Recently, by analyzing HIF-1␣-deficient mouse limbs, Amarilio et al. (52) showed that HIF-1␣ directly binds to and activates the promoter of sox9, a key transcription factor in chondrogenesis. Therefore, hypoxic conditions are necessary to promote the differentiation of prechondrocyte to chondrocytes (52). Stem cells and multipotent progenitor cells reside in complex microenvironments, or niches. During the early stages of development, the vasculature undergoes extensive remodeling. Specifically, oxygen gradients between blood vessels and stem cell niches might serve as important environmental cues to promote or inhibit differentiation, depending on the cell lineage involved (53). Our study provides insight into the role of oxygen, particularly in adipogenesis. Exposure of cells to FIGURE 8. Effect of HIF-1␣ and p27 on GSK-3␤ nuclear localization. A, 3T3-L1 cells were infected with retrovirus encoding p27 Kip1 . As a control, 3T3-L1 cells were infected with retrovirus carrying an empty vector (EV). The infected 3T3-L1 cells were treated with MDI for 24 h under normoxia. Nuclear extracts were isolated from the hormone-treated 3T3-L1 cells, and Western analysis was performed using anti-GSK3␤, anti-C/EBP␤, anti-HIF-1␣, or anti-lamin A/C antibodies. B, preadipocyte 3T3-L1 cells were infected with a retrovirus encoding control shRNA (shCtrl) or shRNAs against mouse HIF-1␣ (shHIF-1␣). The infected 3T3-L1 cells were induced to differentiate by treatment with MDI for 24 h under normoxic or hypoxic conditions for the indicated times. Nuclear extracts were isolated from the hormone-treated 3T3-L1 cells, and Western analysis was performed using anti-GSK3␤, anti-C/EBP␤, anti-HIF-1␣, or anti-lamin A/C antibodies. Lamin A/C was used as a loading control for the nuclear extracts. C, schematic diagram describing the HIF-1␣-dependent and independent mechanisms of hypoxic inhibition of adipogenesis.
hypoxia at any stage of this process can inhibit the progress of adipogenesis by blocking many different checkpoints. Therefore, adipogenesis can proceed only when the vasculature can deliver sufficient oxygen and nutrients.